In just a few hours, ChatGPT can create malicious code that steals data without being detected

According to FOX News , OpenAI’s ChatGPT chatbot has created a lot of buzz in the tech world over the past few months, however not everything this app creates is great. It would be difficult to imagine the damage if ChatGPT was used as a tool to attack the network through malicious software.

The above concern is completely justified when now it only takes a few hours for this application to create a malicious software that helps steal data on the internet.

Fox News quoted an article by security expert Aaron Mulgrew of Forcepoint network security group, saying that he can create malicious software using suggestions on ChatGPT in just a few hours.

According to Mr. Mulgrew, OpenAI has anticipated this risk, so it has built a number of protection measures for ChatGPT, preventing users from asking the application to create malicious software. But this expert still has a way to bypass ChatGPT’s “censorship system”.

Accordingly, Mr. Mulgrew asked ChatGPT to generate functional code with separate requirements. After all the individual functions were recorded, he realized he had a file that could execute the data theft undetectable and as sophisticated as any malware. .

Mulgrew’s test is seen as a warning bell for ChatGPT being weaponized by ordinary users without the need for any advanced hacker groups. Users don’t even have to program malicious software.

The malicious software created by ChatGPT disguised as a screensaver application, then automatically launched on devices using the Windows operating system.

As soon as it appears on the device, it scans all file types including text documents, images and PDFs, and looks for any data it can find to steal from the device. .

After taking hold of the data, it can split the data into smaller pieces and hide those parts in other images on the device.

To avoid detection, stolen images are uploaded to a Google Drive folder. This shows the “smartness” of the software created by ChatGPT.

The entire process of creating malicious software was carried out by Mr. Mulgrew on a separate computer, and the malicious code generated by ChatGPT has not yet hit anyone on the internet.

“I don’t have any advanced coding experience, but the ChatGPT protections are still not strong enough to block my test,” warned Mr. Mulgrew.